A Personal View of Research on Population Impacts on Economic Development
Research on population impacts on economic development is very welcome. The subject is important. And presumably a key concern of the Hewlett/PRB initiative relates to the economic implications (other things equal) of rapid demographic growth.
Of course, the relationship between population growth and economic growth has been a subject of scholarly interest since at least the late eighteenth century. But rapid demographic growth—usually taken to mean a growth rate of 2 percent per year or more—is chiefly a feature of the past sixty years. For example, India's population grew at an average annual rate of about 2 percent between 1951 and 2001, and is now more than triple the size it was in 1947. Many populations in sub-Saharan Africa have grown at rates exceeding 3 percent, with associated doubling times of less than 25 years. According to the UN, for example, Uganda's population in 1950 was about 5 million; today (i.e. in 2005) it is about 29 million; and the medium projection for the year 2050 is about 127 million.
Population growth involving such pace and scale seems very unlikely to be free of challenges for the economy. But this is not the view that is always taken.
In my opinion, the present initiative to get more research conducted on the economic consequences of demographic growth—while appreciated—also raises the question of why more research on the subject by economists has not been undertaken in the first place.
Simple economics suggests that rapid demographic growth is problematic over the long run.1 Other things equal, if a population grows at 2 percent per year, or more, then just to stand still it will have to increase other parameters (e.g. health centres, schools, employment, etc) at about the same rate. Even if its labour force is expanding rapidly as well, there remains the big challenge of converting growth in the labour force into growth in these other parameters (e.g. health centres, schools, etc). It is this sort of idea that lies behind Robert Cassen's telling depiction of rapid population growth as being 'not so much a time bomb as a treadmill'. Of course, in some circumstances there may be economic advantages to a growing population (e.g. economies of scale). But on the other hand there may also be outcomes that are disadvantageous for economic development (e.g. through damage to the resource base).
In the present context, it is important to underscore that many economists have felt that—over the long run, and other things equal—rapid population growth will lead to lower levels of per capita income. Thus some of what follows is a characterization. In addition, many have seen economic advantages to be had from fertility decline—e.g. in terms of feeding and educating children within households. And, at a higher level of aggregation, Coale and Hoover's Population Growth and Economic Development in Low Income Countries—arguably the most important text to be published in the field of population and development in modern times—used simple, but plausible assumptions and simulations to argue that a slower rate of population growth in a poor country like India or Mexico would enable higher rates of investment, and therefore higher levels of per capita income over the longer run. The book was published in 1958 i.e. almost fifty years ago.2
That said, a case can be made that—with some creditable exceptions—the record of scholarship on the issue by academic economists has been rather mixed. For much of the 1980s and 1990s—decades in which the world's population increased at a rate of one billion people every 12 years—demographic growth was arguably neglected and/or downplayed. This was a time of a move towards the right in politics in countries like the US and UK, and research focused on the role of institutions and markets in facilitating economic growth, to the neglect of aspects of population. The views of Julian Simon were pretty dominant during this period. And Simon dismissed Coale and Hoover's book as being merely an 'influential gloss on the basic Malthusian model' (Simon 1989:325). From the 1984 Mexico World Population Conference there came the phrase that 'population growth is, by itself, a neutral phenomenon [with respect to economic development]'. And it must also be said that at the Cairo conference in 1994 the shift towards the individual level (e.g. towards reproductive health) compounded the neglect of more aggregate issues—such as the effect of population growth on living standards.
Research on population impacts on economic development in the 1980s and 1990s involved plotting, for a particular decade, per capita income growth rates against population growth rates. [Of course, there have been various econometric refinements on this, which are more telling]. However, this sort of simple scatter-plot approach was (and remains) central. Indeed, sometimes it appears to be the only thing that counts in economic circles. In such scatter-plots, it is per capita income growth that is—quite rightly—considered as the dependent variable.3
This sort of exercise was often undertaken with data for a huge number of countries, and the scatter-plots were then used to support the conclusion that there was little or no relationship (e.g. see Simon 1989). Thus with regard to such a scatter-plot Barlow (1994:154) remarked that 'the absence of correlation is visually evident, and is confirmed by a linear least-squares regression'. However, it must be remarked that such an approach involves giving equal weight to very different units, e.g. India on the one hand, and comparatively tiny populations like Honduras and Tunisia on the other. It was also unclear whether the same conclusion regarding the absence of a negative relationship held if larger and arguably more comparable units were used instead.4
Beyond these considerations, there are obvious difficulties in drawing dynamic inferences from plots of data that are essentially cross-sectional in nature. Despite acknowledgement of the difficulties of doing this, it is still done a lot (and not just in relation to the main issue under consideration here). This is not to say that cross-sectional scatter-plots don't have their uses. But it is to say that they—and the related drawing of causal inferences from regression coefficients—is often problematic. In any event, it was certainly the case that in the 1980s and 1990s even those scholars who felt strongly that rapid population growth probably did exert a negative influence on economic growth, like Robert Cassen, were forced to choose their words very carefully. For example: 'the available evidence from empirical studies does not show clearly that population growth exerts a negative influence on development' (Cassen 1994). [The key word here is 'clearly'!]
The fact that, with more recent data, a negative relationship between annual population growth rates and rates of per capita income growth has become more evident in such scatter-plots casts doubt on the conclusions drawn from earlier work. Were the units used in the exercises sufficiently comparable? Were the time periods used of sufficient length to draw firm conclusions?
It is worth noting too that around this time a broadly analogous failing occurred with respect to the issue of the potential economic benefits of fertility decline. The World Bank (1993) publication The East Asian Economic Miracle: Economic Growth and Public Policy tried to account for East Asia's rapid economic growth, but it gave rather little attention to the possible role played by fertility decline. This position, of course, has subsequently been challenged—most notably by David Bloom and Geoffrey Williamson (1998) and other economists at Harvard. There remains, however, the basic question of how the issue of demographic change could have been so neglected in the first place.5
The publication, in 2001,of Population Matters: Demographic Change, Economic Growth and Poverty in the Developing World, edited by Nancy Birdsall, Allen Kelley, and Steve Sinding, seemed at last to have put population impacts on economic development onto a more balanced and better plane. Thus in the first chapter, Birdsall and Sinding (2001:6) make two statements based on the work contained in the book. These are that: '[f]irst, in contrast to assessments over the last several decades, rapid population growth is found to have exercised a quantitatively important negative impact on the pace of aggregate economic growth in developing countries … [and] … [s]econdly, [that] rapid fertility decline is found to make a quantitatively relevant contribution to reducing the incidence and severity of poverty.'
These statements seem to be fairly clear and unequivocal. However, there remains doubt as to whether Population Matters does indeed convey a new consensus among economists. For example, citing Kelley's paper in the book, Deepak Lal sees the now prevailing accord in very different—indeed, quite opposite—terms. Thus Lal reads the current consensus among economists as being that: 'the "population problem" is not a problem (except in the very short run, and only if appropriate policies are not in place)'. [One must inject here the observation of Nathan Keyfitz (1991) that '[b]ad policies are widespread and persistent. Social science has to take account of them'. It is questionable of Lal to discount problems simply because 'appropriate polices' will deal with them]. Anyhow, in Deepak Lal's opinion (2006:145) the position of Julian Simon is 'now seen to have been proved right by a burgeoning body of research'.
The editors of, and contributors to Population Matters include many who, like Robert Cassen, have long held the view that rapid population growth hinders economic development, and conversely that fertility decline and slower growth helps it. So economists like Lal might well say that Population Matters is simply an instance of the converted preaching to the converted. It seems that there may not be any new consensus among economists.
The previous paragraphs are tough on (some) economists, and there is a still more to come! A little knowledge—which is what I have on this issue—is a dangerous thing. And it is always hazardous to generalise—particularly when one is being critical. It is important to reiterate that there has been valuable, rounded work on the relationship between population and the economy—not least by economic and demographic historians. Thus there is no better illustration of a negative relationship between the rate of population growth and the rate of change in real wages—i.e. the basic Malthusian model—than the long run data sets assembled for England for the period 1566-1806 (see Wrigley 1988: 34-67).
In fairness, economics is probably no weaker than the other main social sciences—like sociology or political science—in its ability to draw causal inferences. But nor is it obviously any stronger. And nowadays several other social sciences also rely too much on the interpretation of regression coefficients in order to infer causation. They could all learn something—not least with respect to caution—from the principles used to address causation in other fields like epidemiology. Moreover, and of course, demography/population studies (my own area) suffers from similar problems.
Nevertheless, births, deaths, and population change, are hard and discrete events—despite sometime difficulties in measuring them. Economic concepts, on the other hand, tend to be vaguer and they are much harder to gauge in an unequivocal way.
In addition, and not surprisingly, the discipline of economics is limited in that it tends to focus on the 'economic' which, of course, is only one dimension of human life. We know, for example, that mortality decline in the modern world—the main cause of global population growth—owes relatively little to economic growth per se (although rises in living standards do play a part). It is questionable too whether changes in conventional economic variables play much of a role in accounting for the contemporary fertility transition. Indeed, in many ways it is remarkable how marginal economic change has been in accounting for demographic change in the modern world.
Yet another example of how restricted economic research can be in explaining demographic change is afforded by the body of work that, just a couple of decades ago, tried to account for generally higher levels of fertility among women living in landed households compared to landless households in much of Asia. Mead Cain pointed out that this work failed to take any account of the proximate factors conditioning fertility. Thus his fieldwork in Bangladesh suggested that couples in landless households were separated more by seasonal labour migration. And, perhaps more importantly, differences in women's clothing led him to propose that suckling patterns were very different between women in landed and landless households, with potentially considerable implications for fertility differentials. But the theorizing by economists rarely mentioned these things, and was probably invalid as a result.
Few suggestions for future research can be made here that have not been made already by others. Also, this is probably another instance of 'the converted preaching to the converted'. The real task is somewhere 'out there'—in economics departments that have little knowledge of, or interaction with, demography/population studies. Also, there seems to be a need to improve the language of communication; some economists clearly have problems in getting their meaning across to others. Several of the following points are pretty obvious. And in some measure they apply to all of the social sciences.
Countries that face the fastest rates of population growth tend to have the weakest administrative infrastructures. As a result, the quality of their data sets, perhaps especially at the macro-level, is often extremely poor. It may take a very long time for things to improve greatly with respect to large scale data sets in some countries. This needs to be recognised more.
However, for many locations researchers analyse whatever data are available—often with very sophisticated techniques, but with little or no discussion, or appreciation, of data problems. In many ways India's National Sample Survey (NSS) is an outstanding source of information. Its data are analysed in many studies, most notably on poverty and food consumption, but it has also been known for the NSS to collect data on fertility, child mortality and family planning. However, rarely in published papers is it noted that NSS data are collected by male interviewers, usually from male respondents—considerations that surely suggest that some caution is required in interpreting the results.
Most social sciences place great value on theory. And theory is important. Nevertheless, it would be useful if greater emphasis—and value—were given to the need for good basic description. In this context it is worth remarking on the careful work of researchers like Mead Cain back in the 1970s in collecting data on the activities of children. Whether it told us much about why people in poor rural settings had high fertility is doubtful, in my view. It did, however, provide extremely valuable information on what children in rural agricultural societies actually did (see Cain 1977). Time budget studies relating to older people, including the elderly, might be very illuminating.
It is obvious that there is a need for greater knowledge among economists—and other social scientists—about the proximate factors that affect both fertility and mortality. It is still quite common for scholars to address fertility and mortality variation solely with regard to background socio-economic measures. Researchers who have some understanding of demography/population studies are unlikely to commit this mistake. But, unfortunately, they may well be in the minority—beyond the relatively few institutions that teach and have expertise in population studies.
A related point is the need for greater attention to non-economic variables when considering, for example, the impact of population growth on economic development. Here one has in mind especially political and, more still, environmental factors (that may be hard to quantify). For example, population growth may contribute to political instability and/or harm the resource base. And such developments may have deleterious consequences for the economy over the longer run.6 It is worth recalling that Malthus—whose work certainly had breadth—held the first professorship of political economy in the British Isles. Rightly or wrongly, a lot of modern economics has the reputation of being excessively narrow (and quantitative) in focus. Nowadays the scatter-plots of population growth and per capita income growth may indeed be showing what some people always suspected—i.e. a negative relationship. However, there remains a need to extend the range of research well beyond just regression.
More generally, should any social scientist be able to earn a research degree just from sitting at a computer and analysing, say, DHS or Labor Force survey data? The answer may be yes—but only in relation to a small number of subjects. The integration of survey analysis with 'hands on' time in the field should surely become more common. Academic research is often very narrow in focus—concentrating on the 'trees', and missing the 'wood'. Greater combination of survey analysis with fieldwork would help to broaden things and increase understanding.
Some of the complaint expressed above with respect to cross-sectional analysis relates to the lack of attention to time. Conversely, mention of the work on England by Wrigley and others is meant to underscore the importance of time. Research on the consequences of demographic growth in developing countries for their economic development could certainly be more historical. More work of the 'in-depth case study' variety would be valuable.
Also in relation to time, there is a need to pay greater attention to future prospects in a rounded way. A team of us have tried to do this for India (see Dyson, Cassen and Visaria 2004). Starting with state-level population projections for the next quarter century (which are reasonably firm) we then addressed issues relating to the likely broad implications for: urbanization, migration, education, employment, levels of poverty, agriculture, water, energy, the urban environment, politics and administration, and the macro-economy. Notice the prominence of environmental issues in this list. There is not enough work of this integrated, forward looking kind, based around demographic projections. It is probably best done for big units like major world regions (e.g. Europe, sub-Saharan Africa, or West Asia).
If we look at the very considerable population growth that many poor countries—say Yemen, or Mali, or Ethiopia, or Uganda—are set to face in the coming decades, then it seems unlikely that narrow economic research will be of much use. Also, issues of political economy may be very important—for example in relation to migration (e.g. in 2050 will the projected 128 million Ugandans live in Uganda?). It may also be the case that we are increasingly going to have to factor in climate change into any forward-looking analyses.
The recently expressed views of Deepak Lal, reported above, raise the question of the extent to which there is now a consensus regarding the negative effect of population growth on economic development (other things equal). Research into this specific issue might be both valuable and interesting. For example, Oreskes (2004) has conducted similar work with respect to the views of the world's climate scientists on anthropogenic global warming. This work shows conclusively that there is a clear consensus on the issue.
A sociological analysis of economist's views on the relationship between population growth and economic growth in recent decades might also be useful. What I have in mind is essentially an exercise in the 'sociology of knowledge'. As noted, the influence of Julian Simon coincided with and chimed well with contemporaneous political changes in both the US and UK. This was also a time when the World Bank increasingly introduced policies of structural adjustment. However, the case for a study of the work of economists in a sociological vein extends well beyond this. For example, it might be proposed that the views of some showed signs of avoidance and denial. Also, in many fields it is the case that taking an opposite view can be a way of making one's mark.
Future generations may well ask whether, in the era following the second world war, it was morally right to introduce mass death control without giving more attention than we did to the introduction of birth control. Research that sometimes bolstered the view that population growth was generally a neutral phenomenon with respect to economic development (let alone the environment) may have done some real harm. Perhaps the fiftieth anniversary of the publication of Population Growth and Economic Development in Low Income Countries—which, after all, was the result of collaboration between a demographer and an economist—deserves some marking (e.g. perhaps half a dozen oysters and a glass of dry white wine).
Alexandratos, N. (2005). 'Countries with Rapid Population Growth and Resource Constraints: Issues of Food, Agriculture and Development' in Population and Development Review 31(2):237-58.
Barlow, R. (1994). 'Population growth and economic growth: Some more correlations', Population and Development Review 20(1):153-65.
Birdsall, N. and S. Sinding (2001) 'How and why population matters: New findings, new issues', in N. Birdsall, A. C. Kelley, and S. Sinding (eds.) (2001) Population Matters: Demographic Change, Economic Growth, and Poverty in the Developing World, Oxford University Press, Oxford.
Birdsall, N., Kelley A. C. and S. Sinding (eds.) (2001) Population Matters: Demographic Change, Economic Growth, and Poverty in the Developing World, Oxford University Press, Oxford.
Blanchet, D. (1991) 'On interpreting observed relationships between population growth and economic growth: A graphical exposition', Population and Development Review 17(1).
Bloom, D. E. and J. G. Williamson (1998) ‘Demographic transitions and economic miracles in emerging Asia’ World Bank Economic Review 12(3): 419-55.
Cain, M. (1977) 'The economic activities of children in a village in Bangladesh', Population and Development Review 3(3): 201-27.
Cassen, R. (1976) 'Population and Development: A Survey', World Development 4(10/11):785-830.
Cassen, R. (1994) 'Population and development: Old debates, new conclusions', in R. Cassen (ed.) Population and Development: Old Debates, New Conclusions, Washington, DC: Overseas Development Council.
Coale, A. J. and E. Hoover (1958) Population and Economic Development in Low-Income Countries, Princeton, Princeton University Press.
Demeny, P. (1985) 'Bucharest, Mexico City and Beyond', in European Journal of Population 1.
Dyson, T. Cassen, R. and L. Visaria (eds.) (2004) Twenty-first Century India—Population, Economy, Human Development and the Environment, Oxford, Oxford University Press.
Keyfitz N. (1991). Population and development within the ecosphere: One view of the literature. Population Index 57:5-22
Lal. D. (2006) 'India: Population change and its consequences', in P. Demeny and G. McNicoll (eds.) The Political Economy of Global Population Change, Supplement to Population and Development Review 32:145-82.
Livi-Bacci, M. (2001) A Concise History of World Population, Third Edition, Oxford: Blackwell Publishers.
Oreskes, N. (2004) 'The scientific consensus on climate change', Science 306, 5702:1686.
Simon, J. (1989) 'On aggregate empirical studies relating population variables to economic development', Population and Development Review15(2):323-332.
World Bank (1993) The East Asian Economic Miracle: Economic Growth and Public Policy Washington, DC: World Bank.
Wrigley, E. A. (1988) Continuity, Chance and Change, Cambridge, Cambridge University Press.